Google CEO,Eric says

Evolution of Database Technology

1960s:
Data collection, database creation, IMS and network DBMS
1970s:
Relational data model, relational DBMS implementation
1980s:
RDBMS, advanced data models (extended-relational, OO, deductive, etc.)
Application-oriented DBMS (spatial, scientific, engineering, etc.)
1990s:
Data mining, data warehousing, multimedia databases, and Web databases
2000s
Stream data management and mining
Data mining and its applications
Web technology (XML, data integration) and global information systems
 What Is Data Mining?

 Data mining (knowledge discovery from data)

 Extraction of interesting (non-trivial, implicit, previously unknown
and potentially useful) patterns or knowledge from huge amount of
data
 Data mining: a misnomer?
 Alternative names
 Knowledge discovery (mining) in databases (KDD), knowledge
extraction, data/pattern analysis, data archeology, data dredging,
information harvesting, business intelligence, etc.
 Watch out: Is everything “data mining”?
 Simple search and query processing
 (Deductive) expert systems
Why Not Traditional Data Analysis?
Tremendous amount of data
Algorithms must be highly scalable to handle such as tera-bytes of data
High-dimensionality of data
Micro-array may have tens of thousands of dimensions
High complexity of data
Data streams and sensor data
Time-series data, temporal data, sequence data
Structure data, graphs, social networks and multi-linked data
Heterogeneous databases and legacy databases
Spatial, spatiotemporal, multimedia, text and Web data
Software programs, scientific simulations
New and sophisticated applications
 Why Data Mining?

The Explosive Growth of Data: from terabytes to petabytes

Data collection and data availability
Automated data collection tools, database systems, Web, computerized society
Major sources of abundant data
Business: Web, e-commerce, transactions, stocks, …
Science: Remote sensing, bioinformatics, scientific simulation, …
Society and everyone: news, digital cameras, YouTube
We are drowning in data, but starving for knowledge!
“Necessityis the mother of invention”—Data mining—Automated analysis of
massive data sets
Knowledge Discovery (KDD) Process

 Data mining—core of knowledge

discovery process Pattern Evaluation

Data Mining

Task-relevant Data

Data Selection
Warehouse
Data Cleaning

Data Integration

Databases
 KDD Process: Several Key Steps
Learning the application domain
relevant prior knowledge and goals of application
Creating a target data set: data selection
Data cleaning and preprocessing: (may take 60% of effort!)
Data reduction and transformation
Find useful features, dimensionality/variable reduction, invariant representation
Choosing functions of data mining
 summarization, classification, regression, association, clustering
Choosing the mining algorithm(s)
Data mining: search for patterns of interest
Pattern evaluation and knowledge presentation
visualization, transformation, removing redundant patterns, etc.
Use of discovered knowledge
Architecture: Typical Data Mining System

Graphical User Interface

Pattern Evaluation
Knowl
Data Mining Engine edge-
Base
Database or Data
Warehouse Server

data cleaning, integration, and selection

Data World-Wide Other Info

Database Repositories
Warehouse Web
Data Mining and Business Intelligence

Increasing potential
to support
business decisions End User
Decision
Making

Data Presentation Business

Analyst
Visualization Techniques
Data Mining Data
Information Discovery Analyst

Data Exploration
Statistical Summary, Querying, and Reporting

Data Preprocessing/Integration, Data Warehouses

DBA
Data Sources
Paper, Files, Web documents, Scientific experiments, Database Systems
Multi-Dimensional View of Data Mining
 Data to be mined:Relational, data warehouse, transactional, stream,
object-oriented/relational, active, spatial, time-series, text, multi-
media, heterogeneous, legacy, WWW
 Knowledge to be mined: Characterization, discrimination,
association, classification, clustering , trend/deviation, outlier
analysis, etc.
 Techniques utilized: Database-oriented, data warehouse (OLAP),
machine learning, statistics, visualization, etc.
 Applications adapted: Retail, telecommunication, banking, fraud
analysis, bio-data mining, stock market analysis, text mining,
Web mining, etc.
Data Mining: Classification Schemes

 General functionality
 Descriptive data mining
 Predictive data mining
 Different views lead to different classifications
 Data view: Kinds of data to be mined
 Knowledge view: Kinds of knowledge to be discovered
 Method view: Kinds of techniques utilized
 Application view: Kinds of applications adapted
Data Mining: On What Kinds of Data?
Database-oriented data sets and applications
Relational database, data warehouse, transactional database
Advanced data sets and advanced applications
Data streams and sensor data
Time-series data, temporal data, sequence data (incl. bio-sequences)
Structure data, graphs, social networks and multi-linked data
Object-relational databases
Heterogeneous databases and legacy databases
Spatial data and spatiotemporal data
Multimedia database
Text databases
The World-Wide Web
Primitives that Define a Data Mining Task

1. Task-relevant data
Database or data warehouse name
Database tables or data warehouse cubes
Condition for data selection
Relevant attributes or dimensions
Data grouping criteria
2. Type of knowledge to be mined
Characterization, discrimination, association, classification, prediction, clustering,
outlier analysis, other data mining tasks
3. Background knowledge
 A typical kind of background knowledge: Concept hierarchies
 Schema hierarchy :E.g., street < city < province_or_state < country
 Set-grouping hierarchy:E.g., {20-39} = young, {40-59} = middle_aged
4.Pattern interestingness measurements
Simplicity

e.g., (association) rule length, (decision) tree size

Certainty

e.g., confidence, P(A|B) = #(A and B)/ #(B), classification reliability

or accuracy, certainty factor, rule strength, rule quality, discriminating
weight, etc.
Utility

potential usefulness, e.g., support (association), noise threshold

(description)
Novelty

not previously known, surprising (used to remove redundant

5. Visualization/presentation of discovered patterns
Different backgrounds/usages may require different forms of
representation
 E.g., rules, tables, crosstabs, pie/bar chart, etc.
Concept hierarchy is also important
 Discovered knowledge might be more understandable when
represented at high level of abstraction
 Interactive drill up/down, pivoting, slicing and dicing provide
different perspectives to data
Different kinds of knowledge require different representation:
association, classification, clustering, etc.
Data Mining Functionalities
Multidimensional concept description:
1.Characterization and discrimination
Generalize, summarize, and contrast data characteristics, e.g., dry vs. wet
regions
2.Association analysis (correlation vs. causality)
Applyng association rules Frequent patterns,
Diaper  Beer [0.5%, 75%] (Correlation or causality?)
Classification and prediction
Construct models (functions) that describe and distinguish classes or concepts
for future prediction
E.g.,classify countries based on (climate), or classify cars based on (gas
mileage)
Predict some unknown or missing numerical values
 Data Mining Functionalities Contd..
3.Cluster analysis
Class label is unknown: Group data to form new classes, e.g., cluster houses to
find distribution patterns
Maximizing intra-class similarity & minimizing interclass similarity

4. Outlier analysis
Outlier: Data object that does not comply with the general behavior of the data
Noise or exception? Useful in fraud detection, rare events analysis

5.Trend and evolution analysis

Trend and deviation: e.g., regression analysis
Sequential pattern mining: e.g., digital camera  large SD memory
Periodicity analysis
Similarity-based analysis
Other pattern-directed or statistical analyses
Are All the “Discovered” Patterns Interesting?

 Data mining may generate thousands of patterns: Not all of them are interesting
 Suggested approach: Human-centered, query-based, focused mining

 Interestingness measures
 A pattern is interesting if it is easily understood by humans, valid on new or test data
with some degree of certainty, potentially useful, novel, or validates some hypothesis
that a user seeks to confirm

 Objective vs. subjective interestingness measures

 Objective: based on statistics and structures of patterns, e.g., support, confidence,
etc.
 Subjective: based on user’s belief in the data, e.g., unexpectedness, novelty,
actionability, etc.
Find All and Only Interesting Patterns?

 Find all the interesting patterns: Completeness

 Can a data mining system find all the interesting patterns? Do we need to
find all of the interesting patterns?
 Heuristic vs. exhaustive search
 Association vs. classification vs. clustering
 Search for only interesting patterns: An optimization problem
 Can a data mining system find only the interesting patterns?
 Approaches
 First general all the patterns and then filter out the uninteresting ones
 Generate only the interesting patterns—mining query optimization
 Major Issues in Data Mining

1. Mining methodology
Mining different kinds of knowledge from diverse data types, e.g., bio, stream, Web
Performance: efficiency, effectiveness, and scalability
Pattern evaluation: the interestingness problem
Incorporation of background knowledge
Handling noise and incomplete data
Parallel, distributed and incremental mining methods
Integration of the discovered knowledge with existing one: knowledge fusion
 Major Issues in Data Mining
Contd..
2. User interaction
Data mining query languages and ad-hoc mining
Expression and visualization of data mining results
Interactive mining of knowledge at multiple levels of abstraction
3.Applications and social impacts
Domain-specific data mining & invisible data mining
Protection of data security, integrity, and privacy
Potential Applications of DATA Mining

1. Data analysis and decision support

 Market analysis and management
 Target marketing, customer relationship management (CRM), market
basket analysis, cross selling, market segmentation
 Risk analysis and management
 Forecasting, customer retention, improved underwriting, quality control,
competitive analysis
 Fraud detection and detection of unusual patterns (outliers)
2. Other Applications
 Text mining (news group, email, documents) and Web mining
 Stream data mining
 Bioinformatics and bio-data analysis
Ex. 1: Market Analysis and Management
Where does the data come from?—Credit card transactions, loyalty cards, discount coupons, customer
complaint calls, plus (public) lifestyle studies
Target marketing
Find clusters of “model” customers who share the same characteristics: interest, income level, spending
habits, etc.
Determine customer purchasing patterns over time
Cross-market analysis—Find associations/co-relations between product sales, & predict based on such
association
Customer profiling—What types of customers buy what products (clustering or classification)
Customer requirement analysis
Identify the best products for different groups of customers
Predict what factors will attract new customers
Provision of summary information
Multidimensional summary reports
Statistical summary information (data central tendency and variation)
Ex. 2: Corporate Analysis & Risk Management

 Finance planning and asset evaluation

 cash flow analysis and prediction
 contingent claim analysis to evaluate assets
 cross-sectional and time series analysis (financial-ratio, trend analysis, etc.)
 Resource planning
 summarize and compare the resources and spending
 Competition
 monitor competitors and market directions
 group customers into classes and a class-based pricing procedure
 set pricing strategy in a highly competitive market
Ex. 3: Fraud Detection & Mining Unusual Patterns

 Approaches: Clustering & model construction for frauds, outlier analysis

 Applications: Health care, retail, credit card service, telecomm.
 Auto insurance: ring of collisions
 Money laundering: suspicious monetary transactions
 Medical insurance
 Professional patients, ring of doctors, and ring of references
 Unnecessary or correlated screening tests
 Telecommunications: phone-call fraud
 Phone call model: destination of the call, duration, time of day or week. Analyze
patterns that deviate from an expected norm
 Retail industry
 Analysts estimate that 38% of retail shrink is due to dishonest employees
 Anti-terrorism
 Review

 Data mining: Discovering interesting patterns from large amounts of data

 A natural evolution of database technology, in great demand, with wide
applications
 A KDD process includes data cleaning, data integration, data selection,
transformation, data mining, pattern evaluation, and knowledge presentation
 Mining can be performed in a variety of information repositories
 Data mining functionalities: characterization, discrimination, association,
classification, clustering, outlier and trend analysis, etc.
 Data mining systems and architectures
 Major issues in data mining